Knowledge Representation and Reasoning

: (Stanford University - Jaclyn Chen)

- Overview

Knowledge representation and reasoning (KRR) is the study of how to represent information in a way that computers can use to solve problems. KRR is a basic concept in artificial intelligence (AI) research. It is also an important component in many other fields, such as information systems, robotics, and automation.

Here are some definitions of other terms in KRR:

Knowledge representation: The process of representing knowledge in a structured and formal way so that it can be used by computer systems. Knowledge can include objects, events, capabilities, meta-knowledge, facts and knowledge bases.
Logic: The study of correct reasoning, including formal and informal logic. Formal logic is the science of deducing valid inferences or logical truths.
Reasoning: The process of thinking about something logically to form a conclusion or judgment. It can also refer to the mind's ability to think and understand things in a logical way.

Some ways of representing knowledge include:

Enactive: The representation of knowledge through actions
Iconic: The visual summarization of images
Symbolic representation: The use of words and other symbols to describe experiences

KRR is the study of how knowledge can be represented in formal languages and manipulated in an automated way so that computers can make intelligent decisions based on the encoded knowledge. KRR is the part of AI that is concerned with how an agent uses what it knows in deciding what to do.

- Reasoning about Knowledge

Reasoning about knowledge is a method of thinking about knowledge models using logic, deduction, and induction. The goal is to derive new knowledge and understanding of the context.

Reasoning is a thought process that combines two or more thoughts to draw a conclusion to gain new knowledge. Reason is closely linked to logic, which is the deducing of valid conclusions from given starting points or premises.

Reasoning about knowledge is a useful tool in analyzing distributed systems. It is also important to be able to reason about the probability of certain events as well as the knowledge of agents.

- Knowledge Representation in AI

Knowledge Representation in AI describes the representation of knowledge. Basically, it is a study of how the beliefs, intentions, and judgments of an intelligent agent can be expressed suitably for automated reasoning.

Knowledge representation is a fundamental concept in artificial intelligence (AI). It involves creating models and structures to represent information and knowledge in a way that intelligent systems can use.

Knowledge representation and reasoning (KRR) is the field of AI that represents information about the world in a form that a computer system can use to solve complex tasks. For example, KRR can help with diagnosing medical conditions or having a dialog in a natural language.

Some types of knowledge include:

Procedural knowledge: Knowing how to do something, including rules, strategies, procedures, and agendas. Procedural knowledge is often represented as a partial or complete finite-state machine or computer program.
Structural knowledge: A basic problem-solving knowledge that describes the relationship between concepts and objects.
Declarative knowledge: Facts and information about a topic, focusing on the "what" as compared to the "how" or "why". Declarative knowledge is also referred to as verbal or factual knowledge.
Meta knowledge: The knowledge of pre-defined knowledge, including planning, tagging, and learning.

Other types of knowledge representation include:

Simple relational knowledge
Inheritable knowledge
Inferential knowledge

- Four Basic Types of Knowledge Representation

In AI, knowledge can be represented in various ways depending on the structure of the knowledge or the perspective of the designer, or even the type of internal structure used. An effective knowledge representation should be rich enough to contain the knowledge needed to solve the problem. It should be natural, compact and maintainable.

The following are four basic types of knowledge representation techniques:

Logical representation
Semantic network representation (Semantic Web)
Frame representation
Production rules

- Logical Representation

Knowledge and logical reasoning play a huge role in artificial intelligence. However, you often need more than generic and robust methods to ensure intelligent behavior. Formal logic is the most useful tool in this field. It is a language with explicit representation guided by some specific rules. Knowledge representation relies heavily on the logic used, rather than the logical method used to understand or decode knowledge.

It allows the designer to formulate certain important communication rules in order to provide and obtain information from the agent with minimal communication errors. Different logic rules allow you to represent different things, resulting in valid reasoning. Therefore, the knowledge acquired by the logical agent will be deterministic, which means that it is either true or false.

Although working with logical representation is challenging, it forms the foundation of programming languages and enables you to build logical reasoning.

- Semantic Web

Semantic networks allow you to store knowledge in the form of graph networks, where nodes and arcs represent objects and their relationships. It can represent physical objects or concepts or even situations. Semantic networks are often used to represent data or reveal structure. It is also used to support concept editing and navigation.

The Semantic Web is simple and easy to implement and understand. It is more natural than logical representation. It allows you to categorize objects in various forms and then link those objects. It is also more expressive than logical representation.

- Frame Representation

A frame is a collection of properties and their associated values that describe an entity in the real world. It is a record-like structure consisting of slots and their values. Slots can be of different sizes and types. These slots have names and values. Or they could name subfields facets. They allow you to impose constraints on the frame.

There is no limit or limitation on the value of the facets a slot can have, or the number of facets a slot can have, or the number of slots a frame can have. Since a single framework is not very useful, it is more beneficial to build a system of frameworks by collecting interconnected frameworks. It is flexible and can be used by various artificial intelligence applications.

- Production Rules

Representations based on production rules have many properties necessary for knowledge representation. It consists of production rules, working memory, and cycles of recognition behavior. It is also called a condition-action rule. According to the current database, if the condition of the rule is true, the action associated with the rule is executed.

Although production rules lack the precise semantics of rules and are not always valid, rules lead to a higher degree of modularity. It is the most expressive knowledge representation system.

[More to come ...]

Document